vc4: Add support for coalescing ALU ops into tex_[srtb] MOVs.
authorEric Anholt <eric@anholt.net>
Tue, 15 Nov 2016 22:48:43 +0000 (14:48 -0800)
committerEric Anholt <eric@anholt.net>
Tue, 29 Nov 2016 16:52:50 +0000 (08:52 -0800)
commit4690a93b123a64f8730a870a336ae9756d11fd18
tree13ed6e2582fbbfe75ee85c2ef36f0ffbf3f15592
parentf4baf809937c98dabee90ea57b9d7e6135bdb0df
vc4: Add support for coalescing ALU ops into tex_[srtb] MOVs.

This isn't as complete as I would like (can't merge interpolation because
of the implicit r5 dependency, doesn't work with control flow), but this
was cheap and easy.

Improves 3DMMES Taiji performance by 1.15353% +/- 0.299896% (n=29, 16)

total instructions in shared programs: 99810 -> 99059 (-0.75%)
instructions in affected programs:     10705 -> 9954 (-7.02%)
src/gallium/drivers/vc4/vc4_opt_coalesce_ff_writes.c
src/gallium/drivers/vc4/vc4_qir.c
src/gallium/drivers/vc4/vc4_qir.h
src/gallium/drivers/vc4/vc4_qir_emit_uniform_stream_resets.c